AI and I

It was early morning. Still dark outside.

I hooked the battery back up, climbed in, and turned the key.

For a second, nothing. Just that half-breath you hold when you're not sure if weeks of work are about to pay off or fall apart. Then the engine caught. It turned over, settled into an idle, and just... ran.

Better than it did when I bought it.

I sat there for a minute and let that sink in. I had just taken apart the upper intake manifold on a 1991 Nissan Fairlady Z, a job I had no business doing, and put it back together. No formal training. No mechanical background. Just a checklist I'd built with AI, a pile of parts, and a willingness to work through it one step at a time.

That moment changed something in me.

Not because I fixed a car. But because of what it meant: I was now capable of things I had previously written off as beyond me. The ceiling I'd always assumed was there, the one that separates people who build things from people who just use things, turned out to be a lot lower than I thought. And a lot easier to push through.

The Guy Who Always Wanted to Build

I've always been tech-leaning. Always curious. Always drawn to the idea of making something.

In college I was a business major, but I designed my own minor in Business Technology, specifically to sit at the intersection of the two. I took HTML and CSS web design. I got the concepts. But writing real code? Building something from scratch in a programming language? That was a different mountain, and I never climbed it. The syntax, the debugging, the years of practice you needed before you could do anything worth showing. It never clicked. I wanted to make things, not spend years earning the right to try.

So for a long time, I stayed on the consumer side of technology. I used tools. I got good at using tools. But I wasn't building.

That started changing before I even realized it.

The Bot I Built Without Knowing What I Was Building

As an underwriting team lead with eight years in VA mortgage, a big part of my job was answering questions. How does the VA read this scenario? Is this allowable? How do we defend this case? I loved helping my team, but I kept asking myself: what if those answers didn't always have to come from me?

I had a problem I wanted to solve, two free-tier AI tools, and a lot of stubbornness.

So I started building. My goal was a chatbot that could replicate my underwriting knowledge, something you could actually talk to that would check VA guidelines, apply sound underwriting logic, and give you a real answer. I didn't know it at the time, but what I was describing was a RAG system with an agent layer on top. I learned that term after the fact.

The process was anything but clean. I'd hit the message limit on one free tier, copy the code over to the other tool, pick up where I left off, and keep going. Back and forth. Week after week. I had no roadmap. Every session started with "okay, where did we break last time?" Sometimes I lost ground. Sometimes I had to rebuild something I'd already built. But I kept moving.

And then one day it worked. Otto was live. I had shipped something real: an actual system, built from scratch, by someone who couldn't have told you what an API was a few months earlier.

That was the second ceiling that cracked. I didn't set out to become a developer. I set out to solve a problem. The technology met me where I was.

After that, a new rule took hold: not knowing how to do something is no longer a reason not to try.

The Car

The bot gave me confidence. The car gave me proof.

I found documentation and forums online showing the exact job I needed: an upper intake manifold removal on the Z32, also called the plenum, along with a fuel system rebuild underneath. I asked AI to take all of that and organize every step, every part, every tool I'd need, written for someone with zero mechanical training.

Then I worked through the checklist. One item at a time. Lying on a concrete floor at hours I wouldn't normally be awake, second-guessing every torque spec, triple-checking which hose went where. Over the course of a couple of weeks.

And then came that early morning. The battery. The key. The engine turning over and running better than it ever had.

Two projects. Two moments of "I didn't think I could do that." Two new data points that changed what I believe I'm capable of.

The Paper on the Side of the Desk

Back during COVID, my daughter and I were stuck inside looking for things to do. We started playing with dice one afternoon and ended up inventing a math game. You roll two six-sided dice, add the numbers together, and keep rolling until you hit seven. Simple. But it teaches kids to read dice and practice addition in a way that actually feels like play. We called it That's Not Seven.

I thought it was a cool idea, so I wrote it down on a piece of paper.

That piece of paper sat on the side of my desk for years.

Not long ago, I was in my office and tried to toss something into the trash. Missed. Not exactly Kobe. When I leaned down to pick it up off the floor, I saw that piece of paper right there. That's Not Seven.

Before AI, ideas like that just die. You write them down if you're lucky, and then life moves on and takes them with it. What I love most about AI is how it compresses the distance between idea and execution. If you know what levers to pull, that gap gets a lot smaller.

So I took the next step. I worked with Claude to bring that piece of paper to life.

What happened next still gets me. My daughter talked with her school librarian about the game before I even had a working version. She was that proud of it. By the time it was built, the librarian wanted it for the whole school. Now every kid there can play it.

A dice game invented with my daughter during a pandemic, almost thrown in the trash, is live and in classrooms. You can try it yourself.

What I Want You to Take Away

I don't have a computer science degree. I didn't learn to code. I'm not a developer, an engineer, or a data scientist.

I'm someone who got curious, stayed motivated, and stopped letting what I didn't know prevent me from trying.

AI didn't make me smarter. It didn't do the work for me. What it did was meet me where I was and help me move forward: on a car engine, on a chatbot, on a game I invented with my daughter during COVID.

The ceiling you think is there? It's thinner than you believe. And with AI, it gets a little easier to push through every single day.

Try the thing you don't know how to do. Expect to iterate. Expect to fail a few times. Keep going anyway.